the effect that the panel’s report takes a negative view about the potential of
ties with hedonic techniques. Yet our report explicitly concluded: “Hedonics cur-
rently offers the most promising technique for explicitly adjusting observed prices
The issue is not whether hedonics is potentially of great usefulness. It is.
Rather, what is at stake is essentially a choice between two different ways the Bureau
of Labor Statistics could employ its hedonics R&D budget in the near-term future.
The BLS could devote the bulk of those resources toward developing a steady
stream of hedonic equations and incorporating them into the estimation of the
Consumer Price Index. But the panel’s analysis suggests that under current oper-
ating procedures, the results would not be likely to have much effect on the index.
Alternatively, it could, as the panel suggests, channel its efforts principally into
analyses, tests and experiments aimed at exploring and resolving some of the
methodological issues discussed in the panel’s report. The results might well justify
the modification of BLS item replacement procedures and an expanded appli-
cation of hedonics in a way that could make important improvements in the
index.
Going beyond the content of the panel’s report, my own view is that the
research program, among many other goals, could investigate the question of
whether evidence about the pace of technological advance or the market structure
of the industry could be used to predict the degree of coefficient stability. With
some experimentation, it might also be possible to design a regime under which
newly developed hedonic equations would initially be refit at short intervals and the
results used to help determine the appropriate frequency of future refitting. To the
extent that, with sufficient research and experimentation, the Bureau of Labor
Statistics can identify products that are likely to have relatively stable hedonic
coefficients, the current methodology of indirect hedonics can be applied and
expanded with infrequent refitting and reasonable cost. I suspect, however, that the
application of hedonic methodology to sample rotation would require the use of
direct hedonic methodology, which in turn involves continuous refitting of the
equations. The panel recommended that the BLS experiment with the direct
method, beginning with a few carefully selected goods.
As explained earlier, fitting hedonic equations typically requires the expansion
of the current sample of prices collected by the Bureau of Labor Statistics or the
purchase of privately collected data. Under current data collection methodology,
frequent refitting and, even more so, the continuous refitting required by the direct
method could become very costly. This in turn suggests that research on lowering
the costs of data collection through the use of scanners, and perhaps other
techniques, could eventually play an important role in enlarging the scope for
hedonic methods within the Consumer Price Index.
Do'stlaringiz bilan baham: